Retinal Vessel Diameters and the Incidence of Gross Proteinuria and Renal Insufficiency in People With Type 1 Diabetes

The WESDR has been described in detail in earlier publications (24,25). In brief, a probability sample of 10,135 diabetic patients who received primary care in an 11-county area in southern Wisconsin from 1979 to 1980, composed of “younger-onset” and “older-onset” individuals, was invited to participate in the baseline examination conducted from 1980 to 1982 (24,25). These analyses will be limited to the group of younger-onset individuals who were taking insulin and had been diagnosed before 30 years of age (n = 1,210). The second examination was conducted at 4 years (1984–1986) (26), the third at 10 years (1990–1992) (27), the fourth at 14 years (1995–1996) (28), and the fifth at 20 years (2000–2001) from the baseline examination. Reasons for nonparticipation and comparisons between participants and nonparticipants at baseline and follow-up examinations have been presented previously (24–28). All examinations were performed in a mobile examination van in or near the cities where the participants resided, following a similar protocol that was conducted according to the principles expressed in the Declaration of Helsinki, and was approved by the Institutional Human Subjects Committee of the University of Wisconsin.

Serum creatinine was first measured from the second examination (1984–1986) onwards (12). Of the 903 participants who participated in the second examination, we excluded (nonexclusively) those individuals with preexisting renal disease (urine protein level ≥0.3 g/l [n = 195], estimated glomerular filtration rate [GFR] ≤90 ml · min^–1 · 1.73 m⁻² [n = 152], positive history of dialysis [n = 43], or history of renal transplantation [n = 41]); missing serum creatinine levels (n = 36); a history of pancreas transplantation (n = 3); missing information on important covariates: HbA_1c levels [n = 28], BMI [n = 10], alcohol and smoking history [n = 28], and systolic or diastolic blood pressure [n = 30]); and younger than 18 years of age (n = 8). We also excluded those who did not participate in the subsequent examinations and those with urine ketone levels ≥15 mg/dl (n = 29), since the latter could have interfered with serum creatinine measurements (29). This resulted in the inclusion of 557 individuals at risk of incident gross proteinuria and renal insufficiency.

Retinal vessel diameters were measured at the baseline examination using a computer-assisted technique based on a standard protocol and formula for the Atherosclerosis Risk in Communities Study described in detail elsewhere (30). For WESDR, retinal photographs of field 1 at the baseline examination were converted to digital images by a high-resolution scanner using identical settings for all photographs (23). Trained graders masked to participant characteristics measured the diameters of all arterioles and venules coursing through a specified area one-half to one disc diameter surrounding the optic disc using a computer software program (Retinal Analysis; Optimate, Madison, WI). On average, between 7 and 14 arterioles, and an equal number of venules, were measured per eye. Individual arteriolar and venular measurements were combined into summary indexes that reflected the average retinal arteriolar and venular diameter of an eye, based on the Parr-Hubbard formula (30). Reproducibility of this retinal vessel grading approach was high, as previously reported (30).

The assessment and definition of gross proteinuria and renal insufficiency in the WESDR has been previously described (11,12). Semiquantitative determination of urinary protein excretion in casual urine samples was performed using a reagent strip test (Labstix; Ames, Elkhart, IN). Serum creatinine was measured by a method based on a modification of the Jaffe reaction, with a centrifugal analyzer (Roche Cobra FARA; Roche Diagnostics, Division of Hoffmann-La Roche, Nutley, NJ) using standard reagents (Boehringer Mannheim Diagnostic, Indianapolis, IN) (12). The imprecision of the creatinine assay was determined to be 2.73 ± 0.03% (mean ± SD) and 1.45 ± 0.09% on the basis of repeated (n = 140) measurement controls with values of 1.1 and 6.2 mg/dl, respectively. The method was determined to be linear to >20.0 mg/dl. We calculated estimated GFR using the simplified equation developed from the Modification of Diet in Renal Disease study as follows: estimated GFR = 186.3 × (serum creatinine) − 1.154 × age − 0.203 × (0.742 for women) (31,32).

Incident gross proteinuria was defined as the development of ≥0.3 g/l protein excretion in a casual urine specimen (11), and incident renal insufficiency was defined as the development of estimated GFR <60 ml · min^–1 · 1.73 m⁻² (corresponding approximately to a creatinine level >1.5 mg/dl in men or >1.3 mg/dl in women) after the second examination among those at risk (12). In a subsidiary analysis, we also examined associations with incident microalbuminuria (present or absent) estimated from casual urine samples from the second examination onwards.

Protocols for grading diabetic retinopathy have been described in detail elsewhere (24,25) and are modifications of the Early Treatment Diabetic Retinopathy Study adaptation of the Airlie House classification of diabetic retinopathy (33). The retinopathy level for a participant was derived by combining the diabetic retinopathy levels of two eyes. For this study, any retinopathy was defined as level 21 (presence of a single microaneurysm and hard or soft exudates) or higher and proliferative retinopathy was defined as level 60 (new vessels, fibrous proliferation, vitreous or preretinal hemorrhage, or scars of photocoagulation either in scatter or confluent patches) or higher (28).

Trained personnel measured blood pressure according to the Hypertension Detection and Follow-up Program Protocol (34). The mean arterial blood pressure was computed as two-thirds of the diastolic plus one-third of the systolic value. In individuals 25 years of age or older, hypertension was defined as a systolic blood pressure ≥160 mmHg, diastolic blood pressure ≥95 mmHg, and/or a history of antihypertensive medication use at the time of examination; in younger individuals, hypertension was defined as a systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, and/or a history of antihypertensive medication use at the time of examination. Other pertinent procedures included collecting information on combined family income, years of schooling, year of diagnosis of diabetes, cigarette smoking and alcohol intake, standardized measurement of height and weight, and determining HbA_1c and lipid levels from a venous blood sample (24–28,35).

Retinal arteriolar and venular diameter were analyzed as categorical variables (divided into quartiles, with the 1st quartile indicating the smallest and the 4th quartile the largest diameters) and also analyzed as continuous variables (per SD change in arteriolar and venular diameters). We used ANOVA models and χ² tests to compare the relationship of various baseline characteristics in association with retinal venular quartiles. Pooled logistic regression models with increments corresponding to each follow-up examination, similar to time-varying Cox proportional hazards models, were used to determine the RR of gross proteinuria and renal insufficiency, controlling simultaneously for possible confounders (36). The covariates updated with each follow-up examination included age (years), annual family income ($10,000 dollars), HbA_1c level (%), pack-years of smoking, BMI (kg/m²), average alcohol intake (ounces per day), mean arterial blood pressure (mmHg), albuminuria status (categorized into none and microalbuminuria for gross proteinuria as the outcome and none, microalbuminuria, and gross proteinuria for estimated renal insufficiency as the outcome), and retinopathy. For multivariable analysis, we constructed three nested models: the first model was adjusted for age, sex, follow-up time, and annual family income; the second model was further adjusted for duration of diabetes, HbA_1c level, BMI, pack-years of smoking, average alcohol intake, mean arterial blood pressure, and albuminuria status; and the final model was further adjusted for retinopathy levels. Additionally, we tested for any interaction between retinal venular quartiles and follow-up time by including a cross-product interaction term in the logistic regression model; there was no interaction detected (P < 0.20). Finally, we performed stratified analyses within subgroups of several related variables.

Table 1 shows participant characteristics according to retinal venular diameter quartiles. In general, individuals with larger retinal venular diameter were older, more likely to have a longer duration of diabetes and higher BMI, systolic and diastolic blood pressure, and HbA_1c levels. Retinal venular diameter was not significantly related to sex, family income, education, cigarette smoking and alcohol consumption status, or lipid levels. Smaller arteriolar diameter was associated with older age and increased systolic and diastolic blood pressure (data not shown).

After 16 years of follow-up, 183 individuals developed incident gross proteinuria and 114 developed incident renal insufficiency. Larger retinal venular diameter was associated with higher cumulative incidence of gross proteinuria (18.6, 25.4, 37.7, and 50.4%, comparing increasing venular diameter quartiles) and renal insufficiency (10.7, 15.5, 23.2, and 32.8%, comparing increasing venular diameter quartiles). After adjusting for age, sex, follow-up time, and family income, larger retinal venular diameter was associated with an increased risk of gross proteinuria (RR 1.75, 95% CI 1.38–2.21, comparing 4th vs. 1st to 3rd quartiles of venular diameter) and renal insufficiency (1.72, 1.23–2.41) (Table 2). These associations persisted despite controlling for duration of diabetes, HbA_1c levels, blood pressure, baseline retinopathy, and other factors. Retinal arteriolar diameter was not significantly associated with either gross proteinuria or renal insufficiency.

Table 2 also shows these results with retinal venular and arteriolar diameter analyzed as a continuous variable. Each SD increase in retinal venular diameter was significantly associated with a 23% higher risk of gross proteinuria and a 22% higher risk of renal insufficiency, controlling for risk factors. Smaller retinal arteriolar diameter was not associated with increased risk of renal insufficiency.

We performed several stratified analyses to examine potential effect modification for the association of larger venular diameter and risk of gross proteinuria and renal insufficiency (Table 3). The association between larger retinal venular diameter and risk of gross proteinuria and renal insufficiency was seen in individuals with and without microalbuminuria, any retinopathy, or proliferative retinopathy. Associations were also similar in men and women and in groups stratified by cigarette smoking status, BMI, and hypertension status.

Finally, we performed two subsidiary analyses. First, among a subgroup of 323 individuals with information on total cholesterol and HDL cholesterol levels, additional adjustment for these factors showed a similar pattern and strength of associations as those presented in Table 2 (data not shown). Second, we examined the relationship of retinal vessel diameters to incident microalbuminuria defined from a casual urine sample from the second examination onwards. After multivariable adjustment, the incidence of microalbuminuria was associated with larger retinal venular diameter (RR 1.57, 95% CI 1.30–1.90, comparing 4th vs. 1st to 3rd quartiles of venular diameter) but was not related to arteriolar diameter.

In this prospective, population-based study of individuals with type 1 diabetes, larger retinal venular diameter, quantified using a new computer-assisted technique, was associated with risk of diabetic nephropathy developing over a 16-year period. Individuals with retinal venular diameter in the highest quartile of the population were ∼50% more likely to develop gross proteinuria and renal insufficiency than those with venular diameter in the lowest three quartiles, independent of age, sex, duration of diabetes, glycemic control, blood pressure, and baseline retinopathy levels. This association was seen in eyes with and without any retinopathy or proliferative retinopathy. Retinal arteriolar diameter was not associated with diabetic renal disease.

The association of larger venular diameter with incident gross proteinuria probably and renal insufficiency is consistent with other research that links more severe diabetic retinopathy with basement membrane thickening and changes in glomerular extracellular volume in experimental studies (9) and with incident gross proteinuria and renal failure in our previous analysis in the WESDR (11,12). Our data now suggest that retinal venular dilation, independent of more advanced retinopathy, is also related to the development of nephropathy in people with type 1 diabetes.

This finding provides insights into possible mechanisms and pathogenesis of diabetic nephropathy. Cross-sectional data in the WESDR indicated that larger retinal venular diameter was associated with more severe diabetic retinopathy and independently with younger age, higher HbA_1c levels, longer duration of diabetes, higher BMI, and lower mean arterial blood pressure (23). Some of these observations are consistent with studies showing that larger venular diameters and increased blood flow are associated with diabetes status (19,21) and, in people with diabetes, with presence of retinopathy (17,18,20), higher HbA_1c levels (17), and longer duration of disease (22). However, there are no consistent explanations for these observations. One common hypothesis is that venular dilatation is related to hyperperfusion resulting from hyperglycemia (18,21). In eyes with retinopathy, venular dilation has been further hypothesized to be a result of lactic acidosis associated with retinal hypoxia (37). Thus, our finding that larger retinal venular diameter predicts risk of gross proteinuria and renal insufficiency, independent of duration and control of diabetes and retinopathy severity, suggests that microvascular processes resulting from hyperperfusion, ischemia, and other unmeasured factors may precede the development of nephropathy in people with type 1 diabetes.

We did not find a significant association between retinal arteriolar diameter and risk of gross proteinuria or renal insufficiency. Data from the Atherosclerosis Risk in Communities study showed that in the general population, smaller retinal arteriolar diameter was related to chronically elevated blood pressure (38), systemic markers of inflammation (39), and incident diabetes (40), independent of known risk factors. In the WESDR, cross-sectional analysis showed an association between smaller arteriolar diameter and presence of gross proteinuria (23). However, this association was not present in the current prospective analysis. It is possible that individuals with retinal arteriolar narrowing and renal disease are more likely excluded due to selective mortality.

The strengths of our study include a community-based population, long follow-up, quantitative and masked evaluation of retinal vessel diameters, and standardized measurement of renal dysfunction. Study limitations should be highlighted. First, retinal vessel data were available from the baseline study, whereas serum creatinine and gross proteinuria changes were determined from the second examination onwards. Selection bias due to mortality, nonparticipation, and other reasons for loss to follow-up cannot be excluded. For example, people with retinal arteriolar narrowing who developed fatal cardiovascular events may be excluded from the second examination onwards, which may result in nondifferential bias toward the null. Second, the study findings may be influenced by the methods used to collect and measure gross proteinuria (11). For example, factors affected renal hemodynamics (e.g., medications, state of hydration, and dietary protein intake) may lead to false-positive dipstick urine test results. Additionally, use of a single measurement of gross proteinuria may lead to an overestimation of the incidence compared with use of multiple measurements over time. Third, we did not control for variations in pulsatility associated with the cardiac cycle or for other physiological factors. Assuming that the photographs were taken at random during the cardiac cycle, the resulting increase in variability of the grading might further result in an attenuation of our findings. However, variation in cardiac cycle has been estimated to contribute to only 6% of the total variation of arteriolar and venular diameters (41). Finally, it is unclear how these study findings may be applicable to type 2 diabetes and other ethnic groups (as 99% of this cohort was white).

In summary, these population-based data show a prospective association of larger retinal venular diameter and risk of gross proteinuria and renal insufficiency, beyond that attributable to duration and control of diabetes, retinopathy severity, and other risk factors. These data support the concept that early retinal vessel caliber changes precede the development of diabetic nephropathy in people with type 1 diabetes and may provide independent predictive information regarding this complication.

This study was supported by an American Diabetes Association Mentor-based fellowship (to R.K.), National Institutes of Health (Bethesda, MD) Grant no. HL59259 (to R.K. and B.E.K.K.), and in part by Research to Prevent Blindness (New York, NY) (to R.K., Senior Scientific Investigator Award).